GPTalk: Model selection updated  [06.07.26]

The model selection in GPTalk has been updated. A total of 13 AI models are now available: 2 models from Anthropic, 3 models from OpenAI and 8 models provided via GWDG.

GPTalk is available at:
https://gptalk.uni-hohenheim.de/

Current models in GPTalk

Anthropic

Model	Status	Provider	Strengths / typical use cases
Claude Haiku 4.5	unchanged	Anthropic	fast responses, short texts, summaries, simple research and administrative tasks
Claude Sonnet 5	NEW	Anthropic	advanced text work, analysis, structured drafting, longer contexts, strong all-round use

OpenAI

Model	Status	Provider	Strengths / typical use cases
OpenAI GPT 5.4 mini	unchanged	OpenAI	quick everyday tasks, short texts, idea generation, summaries
OpenAI GPT 5.4	NEW	OpenAI	powerful all-round model for text, analysis, concept development, complex tasks and programming support
OpenAI GPT 5.4 nano	unchanged	OpenAI	very fast, simple tasks, short answers, smaller rewrites

GWDG

The GWDG models are provided via GWDG. They come from different model families.

Model	Status	Provider / model family	Strengths / typical use cases
GWDG Gemma 4 31B Instruct	NEW	Google / Gemma	general assistance, text tasks, structured answers, solid open-model alternative
GWDG Mistral Large 3 675B Instruct 2512	unchanged	Mistral AI	complex tasks, analysis, text work, multilingual use
GWDG Qwen 3.5 397B A17B	unchanged	Qwen	advanced prompts, analysis, text work and technical questions
GWDG Qwen 3.6 35B A3B	NEW	Qwen	faster all-round model for standard tasks, text work and technical support
GWDG Qwen 3 Coder Next	NEW	Qwen	programming, code explanation, debugging, scripts and technical documentation
GWDG Qwen 3 Omni 30B A3B Instruct	unchanged	Qwen	versatile use, structured tasks, general assistance
GWDG OpenAI GPT OSS 120B	unchanged	OpenAI	general tasks, text, analysis and technical questions
GWDG GLM 4.7	unchanged	GLM / Zhipu AI	general assistance, text work, analysis, multilingual tasks

Which model should I choose?

For most tasks, a current all-round model such as OpenAI GPT 5.4, Claude Sonnet 5 or one of the larger GWDG models is a good starting point. For short, quick tasks, smaller models such as GPT 5.4 mini, GPT 5.4 nano or Claude Haiku 4.5 are often sufficient. For programming tasks, GWDG Qwen 3 Coder Next is the most obvious choice.

The model choice is not final: depending on the task, it can be worth trying another model. Sometimes the best model is not the largest one, but simply the one that fits the specific question best.

Back to All news